What rework costs and how to calculate it
Rework is the cost that appears on no line of the accounts. How to calculate yours in four steps and decide what to fix first.
Rework is every hour spent fixing something that should already have been finished. It appears on no line of the accounts, because it comes disguised as normal work: the hour of the person redoing it is the same hour as the person doing it, and the system does not tell the two apart.
That is why rework goes unnoticed in companies that watch costs carefully. It is not a new expense, it is an existing expense being consumed twice.
The size of the bill
The quality literature treats this under the name cost of poor quality, or COPQ. Current estimates suggest it can consume between 15% and 20% of an organisation's revenue, counting rework, excessive inspection, delay, wasted capacity and lost customers.
The exact number matters less than the order of magnitude. If your company runs a single-digit net margin and loses a double-digit fraction of revenue to a process that does not get it right first time, rework is not an operational detail: it is the main competitor to your margin.
And there is an aggravating factor the financial sum does not capture. Rework always consumes the same people: the ones who know how to fix things. They are exactly the people who should be working on what does not exist yet.
How to calculate yours, in four steps
Step 1: pick one process, not the whole company
Trying to measure rework across the entire organisation produces a project that never finishes. Pick a process with a clear beginning, middle and end: raising an order, closing a proposal, delivering a service, invoicing. On choosing which process to look at first, read how to find and fix operational bottlenecks.
Step 2: count how many times a month it has to be redone
No system is needed for this. Two or three weeks of simple notes by whoever does the work already gives the order of magnitude. The question worth recording is: is this task being done for the first time, or is it correcting something earlier?
Step 3: estimate the average time of each correction
Include everything the correction drags along: the hour of the person redoing it, the person checking, the person notifying the client, and the person answering the irritated client. The classic error is counting only the execution and ignoring the communication, which usually costs more.
Step 4: multiply and add the indirect effect
| Component | How to estimate it |
|---|---|
| Direct correction hours | Occurrences per month × average time × hourly cost of the person |
| Coordination hours | Time of whoever notifies, chases and follows up |
| Material cost, where it exists | Wasted input, repeated shipping, cancelled invoice |
| Delivery delay | Days late × effect on receipts |
| Lost customers | Frequency of attributable losses × average deal size |
The sum of the first two lines is usually enough to change the conversation. The rest come in when the case calls for them.
The question that saves more than the sum
Once measured, the natural temptation is to create a checking step. That reduces the error reaching the customer and raises the total cost, because checking is work that adds no value, it merely avoids a larger loss.
The question that genuinely resolves it is different: why was the error possible?
A repeated error is almost never carelessness. It is a lack of structure. In the operations I work with, three causes explain most of it:
- The information arrives incomplete at the entrance. If an order can be opened without an essential field, it will be opened without it, and the error is guaranteed further down the line.
- The criterion is not written down. When each person decides by their own judgement, the result varies with whoever was on shift, and half that variation becomes rework.
- The person executing never hears about the error. If the fix happens in another department and nobody tells the source, the cause stays standing, producing the same error indefinitely.
The right automation does not begin with the tool. It begins with clarity about what deserves to be repeated, scaled or eliminated.
Why automating before fixing costs more
Automation applies the existing rule with more speed and more consistency. If the rule is wrong or incomplete, the result is erring faster and more uniformly, which is worse than erring slowly, because the volume grows before anyone notices.
The order that works is known: document, simplify, standardise and only then automate. Simplifying is the step that saves the most and the one most often skipped, because it looks the least technical. It is there that you discover the double check has existed since an incident years ago whose cause was eliminated long since. The full order is in document, simplify, standardise, automate.
Frequently asked questions
How do I identify rework without a management system?
By asking the people doing the work which tasks they repeat most and which depend on checking somebody else's work. Two weeks of simple notes produce a map good enough to prioritise. A system helps with tracking afterwards, it is not a prerequisite for discovering.
What level of rework is acceptable?
It depends on the consequence of the error, not on a universal percentage. In a process where the error is corrected internally before reaching the customer, some rate is economically rational. In a process where the error reaches the customer or has tax implications, the target needs to be far stricter, and investment in prevention pays back quickly.
Is it worth creating a checking step?
As an immediate measure, yes, to contain the damage. As a permanent solution, no: checking is cost that adds no value. It should be treated as a dressing while the cause at the entrance of the process is corrected, and removed afterwards.
How do I get the team to record errors?
By making clear that the measurement is of the process and not of people, and proving it in the first round: if the first record turns into individual blame, there will be no second. Once the team sees that recording led to a process change that eased their own work, the recording sustains itself.
Where this fits in consulting work
Quantifying rework is usually the moment the conversation about technology changes tone. While the discussion is about tools, it is a matter of preference. Once there is a monthly number attached to a specific process, it becomes an investment decision with an estimable return. On building that sum, read how to calculate automation ROI before investing.
If you suspect a meaningful share of your team's capacity is going into fixing, a strategic diagnostic puts that into numbers before any automation proposal.
Sources
Apply
Want to apply this in your business?
A complimentary conversation to understand your context and decide, together, whether it makes sense to move forward.
Marco legal da IA: o que já vale para empresas
O PL 2338 ainda tramita na Câmara, mas parte das obrigações sobre uso de IA já vale hoje via LGPD e ANPD. Veja o que a sua empresa já precisa cumprir.
Client onboarding: how to structure it
A client who starts badly leaves early. How to structure onboarding in four stages, and why it decides retention more than the product does.
Quem decide o quê: a matriz de decisão
Quase toda empresa travada tem o mesmo problema: ninguém sabe quem decide o quê. Como montar a matriz de decisão do negócio em uma única página.